This Survey Report and any recommendations made herein are for the specific facility evaluated and may not be universally applicable. Any recommendations made are not to be considered as final statements of NIOSH policy or of any agency or individual involved. Additional NIOSH Survey Reports are available at http://www.cdc.gov/niosh/surveyreports.

SHORT-TERM MONITORING OF FORMALDEHYDE: COMPARISON OF TWO DIRECT-READING

INSTRUMENTS TO A LABORATORY-BASED METHOD

Deborah V.L. Myers, Ph.D., E.I. Chad H. Dowell, M.S., C.I.H.

and Michael G. Gressel, Ph.D., C.S.P.

NIOSH

W. Dana Flanders, M.D., D.Sc. National Center for Environmental Health

REPORT DATE: June 2009

REPORT NO.: EPHB 331-05b

MANUSCRIPT PREPARED BY: Bernice Clark

U.S. Department of Health and Human Services

Public Health Service Centers for Disease Control and Prevention

National Institute for Occupational safety and Health Division of Applied Research and Technology

4676 Columbia Parkway, MS-R5 Cincinnati, Ohio 45226

2

SITE SURVEYED: Selfield Industrial Park Federal Emergency Management Agency Selma, Alabama SURVEY DATE: July 28–31, 2008 SURVEY CONDUCTED BY: Deborah V.L. Myers, Ph.D, E.I. NIOSH Cincinnati, OH Michael G. Gressel, Ph.D., C.S.P. NIOSH Cincinnati, OH Gary P. Noonan, M.P.A. National Center for Environmental Health/ Division of Environmental Hazards and Health Effects Atlanta, GA Ronald Dobos, C.I.H., C.S.P. Bureau Veritas North America, Inc. Kennesaw, GA William Dendy, REM Bureau Veritas North America, Inc. Kennesaw, GA SITE REPRESENTATIVES CONTACTED: Ronald Parten, Site Manager Selma, AL Randy Brown Selma, AL

3

DISCLAIMER Mention of company names or products does not constitute endorsement by the Centers for Disease Control and Prevention. The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health and the National Center for Environmental Health.

4

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the significant collaboration of CDC and Bureau Veritas

North America for this work. Field guidance, data collection, and data analysis were provided

by Gary Noonan, Liane Hostler, Rick Aspray, Ronald Dobos, William Dendy, Paul Epstein, Sam

Tucker, Dan Farwick, Kevin H. Dunn, Dave Marlow, Brenda Jones, Debbie Fite, Teresa Lewis,

Donald Booher, and Karl Feldmann. Field site assistance was provided by Ronald Parten, Randy

Brown, and other site personnel. Editorial assistance was provided by Ellen Galloway.

5

ABSTRACT

Formaldehyde is used in the production of many household and building products and its health

hazards are well recognized. Airborne formaldehyde concentrations can be measured using

several different techniques, including laboratory-based methods and direct-reading instruments.

For this study, two commercially available direct-reading instruments, an RKI Instruments

Model FP-30 and a PPM Technology Formaldemeter™ htV, were compared with NIOSH Method

2016 in different test environments to determine if these direct-reading instruments can

accurately measure formaldehyde. The RKI Instruments Model FP-30 instrument uses

photoelectric photometry technology to measure formaldehyde, while the PPM Technology

Formaldemeter™ htV instrument uses electrochemical sensing technology. NIOSH Method 2016

is an integrated sampling method that collects formaldehyde on silica gel coated with 2,4-

dinitrophenylhydrazine; the derivitized product (2,4-dinitrophenylhydrazone) is analyzed using

high performance liquid chromatography with UV detection. Forty-seven 1-hour integrated air

samples were collected and analyzed for formaldehyde using NIOSH Method 2016.

Measurements were made simultaneously with both direct-reading instruments and with the

NIOSH Method. The methods yielded the following mean concentrations for the 47 samples:

NIOSH Method 2016, 0.37 ppm; RKI Instruments Model FP-30, 0.29 ppm; and PPM

Technology Formaldemeter™ htV, 0.340 ppm. Pearson correlation showed that the NIOSH

Method and the PPM Technology Formaldemeter™ htV (R2 = 0.902) were more associated than

the NIOSH Method and the RKI Instruments Model FP-30 (R2 = 0.780). Comparison of 1-hour

integrated samples from the three methods showed that on average the RKI Instruments Model

FP-30 instrument (p<0.001) differed significantly from the NIOSH Method 2016, whereas the

PPM Technology Formaldemeter™ htV (p=0.15) was not significantly different from the NIOSH

6

Method. Sensitivity and specificity tests demonstrated that 1-hour integrated samples with the

PPM Technology Formaldemeter™ htV was more accurate at measuring formaldehyde

concentrations greater than 0.2 ppm, while the RKI Instruments Model FP-30 was better at

measuring concentrations less than 0.2 ppm. Although the direct-reading instruments differed

from NIOSH Method 2016, scatter plots and correlation tests showed that the 1-hour integrated

sample collected with the direct-reading instruments correlated with those from the laboratory-

based method.

7

Table of Contents

DISCLAIMER ................................................................................................................................ 3 ACKNOWLEDGEMENTS ............................................................................................................ 4 ABSTRACT .................................................................................................................................... 5 INTRODUCTION .......................................................................................................................... 8 METHODS ..................................................................................................................................... 9

Description of the Direct-Reading Instruments and Laboratory-Based Method ...................... 10 Direct-Reading Instruments ...................................................................................................... 10 Laboratory-Based Method ........................................................................................................ 11 Test Environment ...................................................................................................................... 13 Test Procedure .......................................................................................................................... 14 Data Analysis ............................................................................................................................ 15

RESULTS ..................................................................................................................................... 16 Direct-Reading Instruments versus NIOSH Method 2016 ....................................................... 16 Temperature and Relative Humidity ......................................................................................... 22

DISCUSSION ............................................................................................................................... 22 CONCLUSIONS........................................................................................................................... 25 REFERENCES ............................................................................................................................. 26 APPENDIX ................................................................................................................................... 29

Figures

Figure 1. Scatter Plot of RKI Instruments Model FP-30 versus NIOSH Method 2016 .............. 19 Figure 2. Scatter Plot of PPM Technology Formaldemeter™ htV versus NIOSH Method 2016 . 19 Figure 3. Scatter Plot of PPM Technology Formaldemeter™ htV versus RKI Instruments Model FP-30 ............................................................................................................................................. 20

Tables Table 1. Formaldehyde Method Characteristics .......................................................................... 10 Table 2. Summary Statistics for NIOSH, RKI, and PPM Sampling Methods ............................ 17 Table 3. Correlation Coefficients for NIOSH, RKI, and PPM Sampling Methods ..................... 18 Table 4. Linear Regression Summary of Figures 1, 2, and 3 with Temperature and Relative Humidity Parameters .................................................................................................................... 21

8

INTRODUCTION

Formaldehyde (HCOH) is a colorless, pungent-smelling gas at room temperature [Pickrell et al.

1983]. It is used in the manufacturing of many products, such as plywood, paper, resins,

fertilizers, cosmetics, and medications, and in household products as a preservative [ATSDR

1999; Kelly et al. 1999; Khoder et al. 2000]. Formaldehyde is commonly released into the air by

burning wood or natural gas, from cigarettes or automobiles exhaust, or by off-gassing from

certain materials [U.S. Consumer Product Safety Commission 1997; Khoder et al. 2000].

Formaldehyde is a suspect human carcinogen and a respiratory and skin sensitizer [Khoder et al.

2000]. Formaldehyde concentrations in the environment above the National Institute for

Occupational Safety and Health (NIOSH) recommended exposure limit (REL) of 0.016 parts per

million (ppm), can cause eye, skin, and respiratory tract irritation [NIOSH 2005]. Formaldehyde

use in various building products such as urea formaldehyde foam insulation and wood products

has led to research on its off-gassing rate in mobile homes and other consumer products [Pickrell

et al. 1983].

Airborne formaldehyde can be measured using different techniques, such as laboratory-based

methods and direct-reading instruments. Direct-reading instruments provide results real-time or

within a few minutes; laboratory-based methods, which necessitate sample analysis, can

typically take up to 2 weeks for results. However, comparisons of the reliability and accuracy of

direct-reading methods to a fully evaluated air sampling method for airborne formaldehyde

concentrations are limited. The objective of this study was to compare two direct-reading

formaldehyde instruments in a field application to determine if the direct-reading instruments can

provide accurate measurements. The study compared the RKI Instruments Model FP-30 and the

9

PPM Technology Formaldemeter™ htV with fully evaluated NIOSH Method 2016 [NIOSH

2003]. The null hypothesis for the study was that the mean formaldehyde concentration for each

direct-reading instrument would be equal to the mean concentration measured by NIOSH

Method 2016.

METHODS

This field survey provided an opportunity to compare two commercially available direct-reading

instruments to NIOSH Method 2016 for measuring formaldehyde in air. The direct-reading

instruments used in the comparison study were the RKI Instruments Model FP-30 and the PPM

Technology Formaldemeter™ htV. A brief search was conducted to find formaldehyde direct-

reading instruments. Based on cost and availability, the RKI Instruments Model FP-30 and the

PPM Technology Formaldemeter™ htV were chosen for this study. Details of the three methods

are presented in Table 1 [Peluffo 2009; Roberts 2009]. For tabulated data, the analytical

methods are referred to as NIOSH (NIOSH Method 2016), RKI (RKI Instruments Model FP-30),

PPM (PPM Technology Formaldemeter™ htV), HPLC (high performance liquid

chromatography), UV (Ultraviolet), R2 (coefficient of determination), SD (standard deviation),

SE (standard error), and N (number of observations).

10

Table 1. Formaldehyde Method Characteristics NIOSH RKI PPM Analytical Technique

HPLC, UV detection

Photoelectric photometry

Electrochemical

Range 0.23 to 37 µg/sample

0 to 1.00 ppm 0 to 10.00 ppm

Accuracy ± 19% ± 10% ± 25% Interferants Ozone, ketones,

other aldehydes None reported Phenol, resorcinol,

alcohols, aldehydes, humidity >60%

Flow Rate 0.030 to 1.5 L/min 0.35 L/min 0.353 L/min Sample Period 60 min 15 min 60 sec* *60 seconds in high accuracy mode and 8 seconds in low accuracy mode. For this study, the PPM was in high accuracy mode. Description of the Direct-Reading Instruments and Laboratory-Based Method

The RKI Instruments Model FP-30 uses photoelectric photometry with colorimetric detection

tabs to measure formaldehyde in air (RKI Instruments Inc., Hayward, California) (Figure A-1 in

Appendix). The RKI Instruments Model FP-30 draws in air with an internal pump and uses a

microprocessor to control sample flow rate. The instrument uses detection tabs to sample

airborne formaldehyde. Depending on the desired detection range, a particular tab number is

used during sampling. For this study, number 009 detection tabs were used to measure

formaldehyde concentrations ranging from 0 to 1.00 ppm in a 15-minute sampling time. The

instruction manual states the RKI Instruments Model FP-30 is capable of measuring

formaldehyde concentrations ranging from 0 to 1.00 ppm with a resolution of 0.01 ppm.

According to the manufacturer, measurements are accurate to ± 10% when air temperature is

between 14ºF and 104ºF and the relative humidity is below 90%.

Direct-Reading Instruments

The PPM Technology Formaldemeter™ htV uses electrochemical sensing technology to measure

airborne formaldehyde concentrations (PPM Technology Ltd., Caernarfon, Gwynedd, Wales,

11

United Kingdom) (Figure A-2 in Appendix). The Formaldemeter™ htV collects single air

samples upon initiation by the user, uses a sampling frequency of 1 to 3 minutes, and analyzes

the samples within 60 seconds. The operation manual states the Formaldemeter™ htV is capable

of measuring formaldehyde concentrations ranging from 0 to 10.00 ppm with resolution of 0.001

ppm. Phenol is a positive interference for the Formaldemeter™ htV [PPM Technology Ltd 2005].

To prevent potential sensor interference by phenol, samples were collected with an inline phenol

filter according to the manufacturer’s instructions. The filters can remove phenols from the air

sample without affecting the reading. Alcohols and aldehydes can also cause interference to the

monitor. As noted by the manufacturer [PPM Technology Ltd 2005], the PPM Technology

Formaldemeter™ htV is sensitive to high temperatures and humidity. Relative humidity above

60% could cause a background reading on the instrument. The PPM Technology

Formaldemeter™ htV is specially designed to compensate for relative humidity above 60% using

data analysis functions. Also, the instrument is temperature compensated to operate accurately

in the range of 50ºF–86ºF. Generally formaldehyde concentrations increase with increasing

temperatures in the environment being evaluated [Roberts 2009]. Between the temperatures 50-

86°F, all results fall within ± 25% of the true value at the 95% confidence level for the

instrument, above 86°F the results are still linear (i.e., as temperature increases, the instrument

gives a higher reading) and can be used as a good guideline but the results do deviate from ideal

[Roberts 2009].

NIOSH Method 2016 was used as the reference method for comparison of the two direct-reading

instruments. This method is fully evaluated which included testing to quantify storage stability,

Laboratory-Based Method

12

collection efficiency, and breakthrough volumes [NIOSH 1995]. In order to be fully evaluated, a

NIOSH method must meet a ± 25% accuracy criterion at a 95% confidence level. NIOSH

Method 2016 requires that air samples be collected using a cartridge containing silica gel coated

with 2,4-dinitrophenylhydrazine (2,4-DNPH). For this study, samples were collected on Supelco

S10 LpDNPH cartridges [model #S10L, lot #SP9984]. Because ozone can consume the 2,4-

DNPH reagent and degrade the formaldehyde derivative [NIOSH 2003], a Supelco LpDNPH

ozone scrubber was used with each air sample cartridge. The Supelco LpDNPH ozone scrubber

was connected upstream of the S10L cartridge, which was connected via plastic tubing to the

inlet port of an SKC AirCheck® 2000 sampling pump (SKC Limited, Harrisburg, Pennsylvania)

(pump shown in Figure A-3 in Appendix). All SKC AirCheck® 2000 pumps were calibrated

with a Bios DryCal®

NIOSH Method 2016 laboratory sample results were reported in micrograms (µg) per sample

and parts per billion (ppb). Based on the W-test of log-transformed concentration data, the data

distribution was determined to be lognormal (p=0.961). Therefore, samples with concentrations

below the minimum detectable concentration (MDC) were noted and replaced with the MDC

divided by the square root of 2 [Hornung and Reed 1990]. The MDC is defined as the limit of

DC-Lite (BIOS, Butler, New Jersey) to a flow rate of 0.5 liters per minute

(L/min). Multiple samples were collected with the same pumps throughout the day. Each

collected sample and field blank were immediately capped and placed in individual metallized

packaging to protect the media from air, moisture, and light. The ozone scrubbers were

discarded after each sample. At the end of each sampling day, the samples and field blanks were

shipped on ice (0ºC), according to the method, to the contracting laboratory for analysis. All

samples were analyzed by Bureau Veritas North America, Novi, Michigan. Pre- and post-

sampling pump flow checks were performed at the beginning and end of the sampling day.

13

detection (LOD) divided by the sample volume. The MDC was 2.0 ppb with a 28 L average

sample volume collected over one hour.

For each test location, ambient temperature and relative humidity were measured using a

calibrated VelociCalc Plus monitor (Model 8386A, TSI Incorporated, Shoreview, Minnesota).

This instrument provides real-time temperature and relative humidity measurements.

Test Environment

Results of the Sexton et al. [1989] study suggested manufactured homes and similar structures,

such as temporary housing units (THU) that were purchased by the Federal Emergency

Management Agency (FEMA) for use as temporary housing during national emergency events,

were an appropriate environment for conducting a study comparing two direct-reading

instruments for formaldehyde. For convenience and access to different types of THUs, this study

was conducted at the FEMA storage site at Selfield Industrial Park in Selma, Alabama where a

variety of manufactured homes, park homes, and travel trailers were stored. The three sampling

methods—the two direct-reading instruments and the NIOSH analytical method—were used in

each type of THU and represent a sample set. The different THUs were tested in either

ventilated (windows or doors opened) or unventilated (windows or doors closed) configurations

to provide a wide range of formaldehyde concentrations. The ventilation state was not used as a

variable in data analysis. Providing ventilation decreases formaldehyde concentrations, while

closed units would most likely result in higher concentrations [U.S. Consumer Product Safety

Commission 1997]. A convenience sample of THUs was used for this study. Sampling

14

personnel remained in the THUs for the duration of each sample and wore suitable respiratory

protection.

Test Procedure

All three formaldehyde sampling methods (i.e., sample set) were simultaneously started within

one minute of entering each THU. Ambient temperature and relative humidity measurements

were recorded at the same time with the VelociCalc Plus monitor. All samples were collected

side-by-side in the center of each unit, near or in the kitchen. A 1-hour area air sample for

formaldehyde was collected in each of the THUs in accordance with NIOSH Method 2016 for a

nominal 30 L sample volume. The sampling period for all methods was 1-hour. Because of the

differences in the operation of the instruments, the NIOSH Method 2016, the RKI Instruments

Model FP-30, and the PPM Technology Formaldemeter™ htV collected a different number of

samples during the sampling period. For the direct-reading instruments, these samples were

averaged over the 1-hour sampling period for comparison among the three methods. Three

samples for formaldehyde were collected using the RKI Instruments Model FP-30. Samples

were started every 18 minutes throughout the 1-hour sampling period. The means for the three

RKI Instruments Model FP-30 samples were used in the analysis. The detection tabs were

discarded after each use. Five samples were collected every 12 minutes using the PPM

Technology Formaldemeter™ htV and the means for the five samples were used in the analysis.

The phenol filter was replaced and discarded every fifth sample (as specified in the

manufacturer’s instructions).

15

A standardized sample collection data sheet was used to document relevant sampling data and

notes (Figures A-4 and A-5 in the Appendix). Information included sample start and stop times,

location of sample collection, pump/instrument information, formaldehyde concentrations (in

ppm), and ambient temperature and relative humidity.

Data Analysis

Initially, the data were analyzed using descriptive statistics including means, medians, standard

deviations, and minimum and maximum values. Paired t-tests were used to compare the average

differences among the methods. The association between NIOSH Method 2016 and the two

direct-reading instruments was assessed by bivariate scatter plots, Bland-Altman plots,

correlation analyses, and linear regression. The Spearman and Pearson correlation coefficients

were used as measures of the strength of the association between the formaldehyde values

obtained from the different methods. The sensitivity and specificity of the two direct-reading

instruments were assessed to detect a level above 0.2 ppm by using NIOSH Method 2016 as the

reference concentration method. For the sensitivity and specificity tests, 0.2 ppm was chosen as

the arbitrary cutoff level. The predictive ability of the direct-reading instruments was further

assessed by using each, separately, to predict a level above 0.2 ppm according to NIOSH Method

2016 and calculating the area under the curve. Regression diagnostics included evaluation of

squared terms, use of condition indices to assess potential collinearity, and residual plots. In

sensitivity analyses, results analyzed using a log-transform obtained qualitatively similar results

as the regression diagnostics. All analyses were done in Microsoft Excel software (2003,

16

Microsoft Corporation, Redmond, Washington) and SAS (9.1, SAS Institute Inc., Cary, North

Carolina).

RESULTS

A total of 72 sample sets were collected using each of the three sampling methods (RKI

Instruments Model FP-30, PPM Technology Formaldemeter™ htV, and NIOSH Method 2016).

However, only 47 sample sets were included in the analysis because some of the sampling

pumps used for the NIOSH Method 2016 had pre- and post-sampling flow rate differences

greater than 10%. To address this flow rate problem, only the first and last samples of a

sequence of samples from the same pump with greater than 10% difference in pre- and post-

sampling flow rates were included in the analysis. The concentration for the first sample was

calculated using the pre-sampling flow rate, while the concentration for the last sample was

calculated using the post-sampling flow rate. All other samples with flow rate differences of

10% or less were included in the analysis using the mean of the pre- and post-sampling flow

rates. However, the overall research findings did not change when the mean flow rate value was

applied to the original 72 sample sets (Table A-1 in the Appendix). Results for the direct-

reading instruments were integrated and reported as an average of the three measurements for the

RKI Instruments Model FP-30 and five measurements for the PPM Technology Formaldemeter™

htV.

Direct-Reading Instruments versus NIOSH Method 2016

Table 2 presents descriptive statistics for each sampling method. RKI Instruments Model FP-30

data ranged from 0.01 to 1.00 ppm, the upper limit for the monitor, with a mean of 0.29 ppm.

17

PPM Technology Formaldemeter™ htV data ranged from 0.032 to 1.220 ppm with an average

value of 0.341 ppm. NIOSH Method 2016 data ranged from 0.026 to 1.5 ppm with an average of

0.37 ppm. The RKI Instruments Model FP-30 and NIOSH Method 2016 had a mean difference

of 0.09 ppm (p=0.0007), while the NIOSH Method 2016 and PPM Technology Formaldemeter™

htV had a mean difference of 0.030 ppm (p=0.15). Five sample results from the RKI Instruments

Model FP-30 were recorded as 1.00 ppm, the maximum for the monitor. If these five samples

results from the RKI Instruments Model FP-30 and the associated data points from the NIOSH

Method 2016 and PPM Technology Formaldemeter™ htV are excluded from the analyzed data

set, then the mean values for the 42 observations for the NIOSH Method 2016, RKI Instruments

Model FP-30, and PPM Technology Formaldemeter™ htV methods are 0.29, 0.20, and 0.281

ppm, respectively (Table A-2 in Appendix).

Table 2. Summary Statistics for NIOSH, RKI, and PPM Sampling Methods Method Paired Triplets

N Mean (SD)

(ppm)

Mean Difference vs NIOSH

(SD) (ppm)

Median (ppm)

Min/Max (ppm)

Max-Min Difference

(ppm)

Mean* Squared

Error

NIOSH 47 0.37 (0.35)

— 0.22 0.026 1.5

1.5 —

RKI 47 0.29 (0.32)

0.09 (0.16)

0.07 0.01 1.00

0.99 0.03

PPM 47 0.341 (0.240)

0.030 (0.140)

0.246 0.032 1.220

1.188 0.019

* NIOSH as the reference value for this analysis The Spearman correlation was used to test the nonparametric correlation between the direct-

reading instruments and NIOSH Method 2016, while the Pearson correlation coefficient was

used to measure the linear relationship between the methods. Table 3 shows the correlation

coefficients for the different methods. The Pearson correlation coefficient for the RKI

18

Instruments Model FP-30 and NIOSH Method 2016 was 0.780, and 0.902 for the PPM

Technology Formaldemeter™ htV and NIOSH Method 2016. The Spearman correlation

coefficient for the RKI Instruments Model FP-30 and NIOSH Method 2016 was 0.883, and

0.949 for the PPM Technology Formaldemeter™ htV and NIOSH Method 2016. Figure 1

presents a scatter plot and the regression line for the RKI instrument versus NIOSH Method

2016. Figure 2 presents a scatter plot and regression line for the PPM instrument versus NIOSH

Method 2016. Figure 3 shows a scatter plot and regression line for the two direct-reading

instruments.

Table 3. Correlation Coefficients for NIOSH, RKI, and PPM Sampling Methods Method NIOSH

Pearson/Spearman RKI

Pearson/Spearman PPM

Pearson/Spearman NIOSH 1 / 1 0.780 / 0.883 0.902 / 0.949 RKI 0.780 / 0.883 1 / 1 0.699 / 0.836 PPM 0.902 / 0.949 0.699 / 0.836 1 / 1

19

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1 1.2

NIOSH Method 2016 (ppm)

RK

I Ins

trum

ents

Mod

el F

P-30

(ppm

)

Figure 1. Scatter Plot of RKI Instruments Model FP-30 versus NIOSH Method 2016

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1 1.2

NIOSH Method 2016 (ppm)

PPM

Tec

hnol

ogy

Form

alde

met

er™

htV

(ppm

)

Figure 2. Scatter Plot of PPM Technology Formaldemeter™ htV versus NIOSH Method 2016

20

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1 1.2

RKI Instruments Model FP-30 (ppm)

PPM

Tec

hnol

ogy

Form

alde

met

er™

htV

(ppm

)

Figure 3. Scatter Plot of PPM Technology Formaldemeter™ htV versus RKI Instruments Model FP-30 Table 4 displays the multiple linear regression summary of the three methods with additional

parameters, temperature and relative humidity. Temperature and relative humidity were added to

the model to determine if the parameters influenced the relationship between the NIOSH Method

and each of the two direct-reading instruments. Bland-Altman plots were done with all methods

(Figures A-6 through A-8 in the Appendix) to measure if any two of the methods were

correlated.

21

Table 4. Linear Regression Summary of Figures 1, 2, and 3 with Temperature and Relative Humidity Parameters Univariate

Slope (SD) Univariate Model R2

Multivariate‡

Slope (SE) Multivariate‡

Model R2

NIOSH vs RKI* Intercept (model w/RKI)

0.098 (0.032) −2.643 (1.067)

RKI 0.957 (0.076) 0.780 0.884 (0.080) Temperature −0.031 (0.007) 0.280 −0.017 (0.008) Relative Humidity

0.028 (0.005) 0.348 0.019 (0.006)

Error Variance 0.022 0.829 NIOSH vs PPM* Intercept (model w/ PPM)

0.0959 (0.028) 0.643 (0.266)

PPM 1.368 (0.067) 0.902 1.282 (0.069) Temperature† −0.031 (0.007) 0.280 −0.008 (0.003) — — — Error Variance 0.010 0.917 *NIOSH as dependent variable, RKI and PPM as independent variables †Relative humidity omitted, as not significant, minor improvement in R2 (0.919) ‡In all multivariate models, temperature and relative humidity showed evidence of collinearity with the intercept. For the purpose of this study, sensitivity and specificity were defined with an arbitrary value.

The observations were concentrated at and below 0.2 ppm on the scatter plots; therefore, 0.2 ppm

was picked as the arbitrary value for the sensitivity and specificity calculations. The sensitivity

is the probability of a true positive value above 0.2 ppm, whereas the specificity is the

probability of a false positive value below 0.2 ppm. For detecting a formaldehyde level above

0.2 ppm according to NIOSH Method 2016, the RKI Instruments Model FP-30 had 60%

sensitivity and 92% specificity. The receiver operating characteristic (ROC) curve is a plot of

the true positive values against the false positive values. The area under the ROC curve based on

logistic regression was 0.886 for the RKI Instruments Model FP-30. For the PPM Technology

Formaldemeter™ htV to detect a formaldehyde level above 0.2 ppm, the sensitivity was 88% and

22

the specificity was 65%. The area under the ROC curve based on logistic regression was 0.910

for the PPM Technology Formaldemeter™ htV.

Temperature and Relative Humidity

Temperatures in the THUs ranged from 80ºF–102ºF with an average temperature of 89ºF.

Relative humidity ranged from 50%–83% with an average reading of 67%. Relative humidity

was positively associated with formaldehyde concentrations for all three methods, whereas

ambient temperature was negatively associated with the methods.

DISCUSSION

The purpose of this study was to compare two commercially available direct-reading instruments

with NIOSH Method 2016. The means of the integrated formaldehyde measurements obtained

using the direct-reading instruments were positively correlated with measurements obtained

using NIOSH Method 2016 (R2 = 0.780 for RKI Instruments Model FP-30 and 0.902 for PPM

Technology Formaldemeter™ htV). However, the RKI Instruments Model FP-30 method

(p<0.001) yielded statistically significantly lower readings than NIOSH Method 2016, whereas

the differences, on average with the PPM Technology Formaldemeter™ htV (p=0.15) were not

statistically significantly different from NIOSH Method 2016.

All three scatter plots showed a positive correlation between the different types of formaldehyde

measurements. Both direct-reading instruments had a higher positive correlation with the

NIOSH Method (R2 = 0.780 for RKI Instruments Model FP-30 and 0.902 for PPM Technology

23

Formaldemeter™ htV) than with each other (R2 = 0.699). The Bland-Altman plots suggested

little pattern in the differences between the three methods other than possibly greater variation at

higher values.

Approximately 78% of the variability in the RKI Instruments Model FP-30 results is accounted

for by a linear relationship with NIOSH Method 2016 sample collection results (Table 2).

According to the manufacturer, the high temperature and relative humidity readings from this

study should not have caused a background reading on the instrument [RKI Instruments 2003].

Of the variability in the PPM Technology Formaldemeter™ htV results, 90% is accounted for by

a linear relationship with NIOSH Method 2016 data (Table 2). The PPM Technology

Formaldemeter™ htV was calibrated each day in the morning when temperatures and relative

humidity were within the recommended parameters. Because both temperature (89ºF) and

relative humidity (67%) were above the favorable operating range, the reliability of the PPM

instrument’s formaldehyde readings was a concern with higher formaldehyde readings than

reported by the monitor.

For this study, temperature was no longer significant, after accounting for relative humidity’s

correlation to NIOSH Method 2016. There was a negative association of NIOSH Method 2016

with temperature (e.g., as temperature increased, NIOSH Method 2016 formaldehyde readings

decreased). This negative association could have been caused by the ventilated condition as the

temperature inside the THUs increased. As the temperature in the THUs increased, the doors

and windows of the THUs were opened. This change in ventilation state may have confounded

24

with the temperatures in the THUs causing the temperature to have no importance as a parameter

in data analysis.

Sensitivity and specificity tests showed that the PPM Technology Formaldemeter™ htV was

better at predicting formaldehyde concentrations greater than 0.2 ppm compared to the RKI

Instruments Model FP-30. Temperature and ventilation state may have affected each

instrument’s ability to detect formaldehyde levels greater than the arbitrary value of 0.2 ppm.

According to the manufacturer, the PPM Technology Formaldemeter™ htV monitor requires

regular calibration to guarantee the monitor is functioning properly. The formaldehyde

calibration standard, thermometer, and temperature/concentration table are used to check and

adjust the calibration of the instrument. However, the thermometer supplied with the PPM

Technology Formaldemeter™ htV monitor is difficult to use and could affect the accuracy of the

monitor’s calibration. Accurate readings from the thermometer are needed because

formaldehyde concentrations in the calibration standard vary with temperature [PPM Technology

Ltd 2005]. Additionally, the temperature/concentration table is only useful for temperatures in

the range of 59ºF–84ºF. Because the highest temperatures in this field study were outside the

recommended range, the readings from the PPM Technology Formaldemeter™ htV monitor may

not have been accurate. This temperature difference was a weakness for this field study because,

according to the temperature/concentration table, a one degree change in temperature means

more than a 10% difference in concentration.

25

One potential drawback of the RKI Instruments Model FP-30 monitor is its inability to report

formaldehyde concentrations greater than 1.00 ppm. In some applications this could be a major

issue, but for screening to detect elevated levels (e.g., greater than 0.2 ppm), this should pose

little or no problem. The monitor gives a reading of “over,” if the concentration measured is

greater than 1.00 ppm. The five RKI Instruments Model FP-30 readings that were above the

monitor’s limit would have been greater than 1.00 ppm if a value had been reported.

Other weaknesses of this field study included the lack of precision (the instruments may have

varied independently of the variation in the NIOSH reference method). A reference method is a

benchmark and research standards preclude the altering of the benchmark, which was another

weakness in the study. Therefore, the NIOSH Method 2016 could not be manipulated in this

study. Not being able to use 25 of the 72 sample results in the analysis was another weakness

since it decreased the sample size and therefore the ability to detect statistical differences. Pump

calibration issues and the operation of PPM Technology Formaldemeter™ htV instrument above

its recommended temperature range were other weaknesses in this study. Despite the

weaknesses, the field study provided a realistic test environment and flexibility to compare the

different methods.

CONCLUSIONS

The 1-hour integrated sample collected with the RKI Instruments Model FP-30 was statistically

significantly different from the NIOSH Method 2016 (p<0.001), whereas the 1-hour integrated

sample collected with the PPM Technology Formaldemeter™ htV was not statistically

significantly different from the NIOSH Method (p=0.15). High temperature and relative

humidity readings, the direct-reading instruments’ operating capability and environment

sensitivity, and pump pre- and post-sampling flow rate differences greater than 10% may have

26

affected the relationship between the instruments and NIOSH Method 2016. Under the test

situations, the PPM Technology Formaldemeter™ htV had similar, though slightly better,

discriminatory ability for detecting formaldehyde concentrations over 0.2 ppm compared to the

RKI Instruments Model FP-30, based on the area under the ROC curve and a 1-hour integrated

sample.

Although the direct-reading instruments differed from NIOSH Method 2016, scatter plots and

correlation tests showed that the 1-hour integrated sample collected with the instruments

correlated with those from the laboratory-based method. A 1-hour integrated sample collected

with the direct-reading instruments maybe useful as screening tools and might preclude the need

for timely and expensive laboratory analysis if low concentrations are measured. However,

additional field evaluations under a variety of environmental conditions and formaldehyde

concentrations are needed to better understand the accuracy of these direct-reading instruments

as compared to NIOSH Method 2016.

REFERENCES

ATSDR [1999]. Agency for Toxic Substances and Disease Registry toxicological profile for

formaldehyde. Atlanta, GA: U.S. Department of Health and Human Services, Centers for

Disease Control and Prevention, Agency for Toxic Substances and Disease Registry.

[www.atsdr.cdc.gov/toxprofiles/tp111.html]. Date accessed: May 2009.

Hornung RW, Reed LD [1990]. Estimation of average concentration in the presence of

nondetectable values. Appl Occup Environ Hyg 5(1):46–51.

27

Kelly TJ, Smith DL, Satola J [1999]. Emission rates of formaldehyde from materials and

consumer products found in California homes. Environ Sci Tech 33(1):81–88.

Khoder MI, Shakour AA, Farag SA, Abdel Hameed AA [2000]. Indoor and outdoor

formaldehyde concentrations in homes in residential areas in Greater Cairo. J Environ Monit

2(2):123–126.

NIOSH [1995]. Guidelines for air sampling and analytical method development and evaluation.

Washington, DC: U.S. Department of Health and Human Services, Centers for Disease Control

and Prevention, National Institute for Occupational Safety and Health, DHHS (NIOSH)

Publication No. 95-117. [www.cdc.gov/niosh/docs/95-117/]. Date accessed: May 2009.

NIOSH [2003]. Formaldehyde: Method 2016 (supplement issued 01/15/1998). In: Eller PM,

Cassinelli ME, eds. NIOSH manual of analytical methods. 4th ed. Cincinnati, OH: U.S.

Department of Health and Human Services, Centers for Disease Control and Prevention,

National Institute for Occupational Safety and Health, DHHS (NIOSH) Publication No. 94-113.

[www.cdc.gov/niosh/nmam/]. Date accessed: May 2009.

NIOSH [2005]. Pocket guide to chemical hazards. Cincinnati, OH: U.S. Department of Health

and Human Services, Centers for Disease Control and Prevention, National Institute for

Occupational Safety and Health, DHHS (NIOSH) Publication No. 2005-149.

[www.cdc.gov/niosh/npg/]. Date accessed: May 2009.

28

Peluffo S (speluffo@rkiinstruments.com) [2009]. RKI Tech Support Web Inquiry. Private e-

mail message to RKI Instruments, Inc. (support@rkiinstruments.com), April 22.

Pickrell JA, Mokier BV, Griffis LC, Hobbs CH [1983]. Formaldehyde release rate coefficients

from selected consumer products. Environ Sci Tech 17(12):753–757.

PPM Technology Ltd [2005]. PPM Formaldemeter™ htV 3 parameter IAQ monitor operation

manual. Caernarfon: Wales, UK.

RKI Instruments Inc [2003]. Instruction manual for formaldehyde gas detector model FP-30/FP-

40. Hayward, CA.

Roberts L (laura@ppm-technology.com) [2009]. Formaldemeter questions. Private e-mail

message to PPM Technology Ltd (technical@ppm-technology.com), April 22.

Sexton K, Petreas MX, Liu K [1989]. Formaldehyde exposures inside mobile homes. Am Chem

Soc 23(8):985–988.

U.S. Consumer Product Safety Commission [1997]. An update on formaldehyde-1997 revision.

[www.cpsc.gov/cpscpub/pubs/725.html]. Date accessed: May 2009.

29

APPENDIX

Figure A-1. RKI Instruments Model FP-30

30

Figure A-2. PPM Technology Formaldemeter™ htV

31

Figure A-3. SKC AirCheck® 2000

32

Figure A-4. Sample Data Sheet Page 1

33

Figure A-5. Sample Data Sheet Page 2

34

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0 0.2 0.4 0.6 0.8 1 1.2 1.4

Mean of Both Methods (ppm)

Diff

eren

ces B

etw

een

Met

hods

(ppm

)

Figure A-6. Mean of NIOSH Method 2016 and RKI Instruments Model FP-30. Bland-Altman plot of the data obtained from 47 paired samples measured with NIOSH Method 2016 and RKI Instruments Model FP-30. Correlation R = 0.3898 (p<0.01). Slope = 0.1553 (p<0.01). Intercept = 0.0795 (p<0.01).

35

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0 0.2 0.4 0.6 0.8 1 1.2 1.4

Mean of Both Methods (ppm)

Diff

eren

ces B

etw

een

Met

hods

(ppm

)

Figure A-7. Mean of NIOSH Method 2016 and PPM Technology Formaldemeter™ htV. Bland-Altman plot of the data obtained from 47 paired samples measured with NIOSH Method 2016 and PPM Technology Formaldemeter™ htV. Correlation R = 0.6798 (p<0.01). Slope = 0.2363 (p<0.01). Intercept = 0.0163 (p=0.35).

36

-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

0 0.2 0.4 0.6 0.8 1 1.2 1.4

Mean of Both Methods (ppm)

Diff

eren

ces B

etw

een

Met

hods

(ppm

)

Figure A-8. Mean RKI Instruments Model FP-30 and PPM Technology Formaldemeter™ htV. Bland-Altman plot of the data obtained from 47 paired samples measured with RKI Instruments Model FP-30 and PPM Technology Formaldemeter™ htV. Correlation R = 0.4508 (p<0.01). Slope = 0.1710 (p<0.01). Intercept = 0.1005 (p<0.01).

37

Table A-1. Summary Statistics for NIOSH, RKI, and PPM Sampling Methods Method Paired Triplets

N Mean (SD) (ppm) Mean Difference vs NIOSH (SD)

(ppm)

Min / Max (ppm)

NIOSH 72 0.37 (0.34) — 0.0014 / 1.5 RKI 72 0.26 (0.29) 0.11 (0.050) 0.013 / 1.00 PPM 72 0.349 (0.250) 0.0210 (0.0900) 0.0320 / 1.270 Table A- 2. Summary Statistics for NIOSH, RKI, and PPM Sampling Methods for Sample Results Greater than 1 Method Paired Triplets

N Mean (SD) (ppm) Mean Difference vs NIOSH (SD)

(ppm)

Min / Max (ppm)

NIOSH 5 1.1 (0.29) — 0.77 / 1.5 RKI 5 1.00 (0.00) 0.10 (0.29) 1.00 / 1.00 PPM 5 0.843 (0.222) 0.257 (0.068) 0.664 / 1.220